Supplementary Information - Nature Research...37005 Ceske Budejovice and Institute of Nanobiology...
Transcript of Supplementary Information - Nature Research...37005 Ceske Budejovice and Institute of Nanobiology...
1
Supplementary Information
Dynamics and hydration explain failed functional transformation in
dehalogenase design
Jan Sykora1†
, Jan Brezovsky2†
, Tana Koudelakova2†
, Maryna Lahoda3, Andrea Fortova
2,
Tatsiana Chernovets1, Radka Chaloupkova
2, Veronika Stepankova
2, Zbynek Prokop
2,4, Ivana
Kuta Smatanova3, Martin Hof
1*, Jiri Damborsky
2,4*
1 J. Heyrovský Institute of Physical Chemistry ASCR, v. v. i., Dolejškova 3, 182 23 Prague 8,
Czech Republic
2 Loschmidt Laboratories, Department of Experimental Biology and Research Centre for Toxic
Compounds in the Environment RECETOX, Faculty of Science, Masaryk University,
Kamenice 5/A13, 625 00 Brno, Czech Republic
3 Faculty of Science, University of South Bohemia in Ceske Budejovice, Branisovska 31,
37005 Ceske Budejovice and Institute of Nanobiology and Structural Biology ASCR, Zamek
136, 37333 Nove Hrady, Czech Republic
4 International Clinical Research Center, St. Anne's University Hospital Brno, Pekarska 53, 656
91 Brno, Czech Republic
†These authors contributed equally to this work
*To whom correspondence should be addressed. E-mail: [email protected],
Nature Chemical Biology: doi:10.1038/nchembio.1502
2
Table of contents
Supplementary note .................................................................................................................... 3
Molecular modeling .................................................................................................................. 3
Supplementary results ................................................................................................................ 6
Supplementary Figure 1. Circular dichroism spectra of studied enzymes................................ 6
Supplementary Figure 2. Comparison of the residues involved in the transplantation. ........... 7
Supplementary Figure 3. Superposition of the catalytic pentads. ............................................. 8
Supplementary Figure 4. MALDI-TOF MS spectra of DhaA12 and DhaA12-coumarin
complex. .................................................................................................................................... 9
Supplementary Figure 5. Acrylamide quenching of coumarin fluorescence for the haloalkane
dehalogenase variants. ............................................................................................................ 10
Supplementary Figure 6. Structural formula of the fluorescent coumarin probe. .................. 11
Supplementary Figure 7. Two lowest-energy conformations of the coumarin probe. ........... 12
Supplementary Figure 8. Three lowest-energy conformations of the covalent linker. ........... 13
Supplementary Figure 9. Five reactive binding modes of the probe bound in DbjA. ............ 14
Supplementary Figure 10. Three reactive binding modes of the probe bound in DhaA. ....... 15
Supplementary Figure 11. Five reactive binding modes of the probe bound in DhaA12. ..... 16
Supplementary Table 1. Characteristics of enzymes used in this study. ................................ 17
Supplementary Table 2. Enantioselectivity of wild type and variants with bromoalkanes
and bromoester. .................................................................................................................. 18
Supplementary Table 3. Data collection and refinement ........................................................ 19
statistics for the crystal structure of DhaA12. ......................................................................... 19
Supplementary Table 4. Parameters characterizing the overall time-dependent fluorescence
shift. ........................................................................................................................................ 20
Supplementary Table 5. Parameters describing the dynamics and hydration of wild types and
DhaA12 mutant. ...................................................................................................................... 21
Supplementary Table 6. Variable residues in the second and the third shell of the active sites
of DbjA and DhaA12. ............................................................................................................. 22
Supplementary Table 7. Atom types and charges for the fluorescent probe used in the
simulations. ............................................................................................................................. 23
Supplementary Table 8. Atom types and charges for the covalent linker bound to the aspartic
acid residue used in the simulations. ...................................................................................... 24
Supplementary Table 9. Binding energies of individual binding modes of the fluorescent
probe. ...................................................................................................................................... 25
Supplementary references ....................................................................................................... 26
Nature Chemical Biology: doi:10.1038/nchembio.1502
3
Supplementary note
Molecular modeling
Preparation of structures for molecular modeling. Crystal structures of the haloalkane
dehalogenases were obtained from RSCB PDB database using following accession codes:
1CQW (DhaA), 3SK0 (DhaA12) and 3A2M (DbjA). The catalytic histidine was substituted by
phenylalanine in all protein structures and the mutations Val172Ala, Ile209Leu and Gly292Ala
were modeled to the crystal structure of DhaA to achieve correspondence with the
experimentally used haloalkane dehalogenase DhaA from the strain NCIMB13064
(DDBJ/GenBank/EMBL accession no. AF060871). All mutations were designed in
PyMOL 1.4.1 (Schrödinger). Hydrogen atoms were added to the protein structures using H++
server at pH 7.51.
Molecular docking. The fluorescent probe consisting of coumarin 120 dye carrying
covalent alkyl-halogen linker (Supplementary Fig. 6a) was built in PyMOL 1.4.1. The
geometry of probe molecule was optimized using AM1 semi-empirical quantum chemical
method2 implemented in the program MOPAC2009 (Stewart Computational Chemistry). The
structures of the enzymes and the fluorescent probe were prepared for docking calculation by
MGLTools 1.5.03. The fluorescent probe was docked into the active sites of all three studied
enzymes using the program AUTODOCK 3.054. The atomic and electrostatic maps were
calculated using AUTOGRID 3.064. Range of grid maps was set to 95 95 95 grid points
with spacing 0.25 Å centered on CG atom of substituted phenylalanine to cover the whole
active site and the main tunnel of both enzymes. In total, 250 docking calculations were
performed employing the Lamarckian Genetic Algorithm with the following parameters: initial
population size 150, maximum of 1.5 106 energy evaluations and 27,000 generations, elitism
value 1, mutation rate 0.02 and cross-over rate 0.8. The local search was based on the pseudo
Solis and Wets algorithm with a maximum of 300 iterations per local search5. The final
orientations from each docking were clustered with a clustering tolerance for the
root-mean-square positional deviation of 2 Å.
Parameterization of modified amino acid residue for molecular dynamics simulation. The
novel residue representing the fluorescent probe covalently bound to the nucleophile of an
aspartic acid (Supplementary Fig. 6b) was parameterized by analogy with the Cornell et al.
force field6. The only exception was the aromatic oxygen that was present in the moiety of
fluorescent probe. The required parameters were obtained from the VanBeek et al7. The novel
residue was constructed from two fragments (i) the coumarin probe capped with
N-methylamide (NME) residue (Supplementary Fig. 7), and (ii) the covalent linker capped with
acetyl (ACE) residue covalently bound to the aspartic acid residue capped with NME and ACE
Nature Chemical Biology: doi:10.1038/nchembio.1502
4
residues (Supplementary Fig. 8). The coumarin fragment was modeled in two lowest-energy
conformations (Supplementary Fig. 7), while the second fragment was modeled in three
lowest-energy conformations keeping the helical conformation of the aspartic acid residue
(Supplementary Fig. 8). The geometry of all lowest-energy conformations were energy
minimized with MP2/6-31G* wave function using Gaussian09 program revision C.01
(Gaussian, 2009). The partial atomic charges of the novel residue were obtained using RESP
ESP charge Derive 2.0 server8,9
, with HF/6-31G* level of theory. The charges on the fragments
were developed employing RESP-A1A charge model, using multi-conformation
multi-orientation RESP fits. During the fitting procedure, the charges on the capping NME and
ACE residues were constrained to zero. The charges on the four peptide bond atoms
(N43, H44, C45 and O46) of the second fragment were constrained to the values corresponding
to the charges on respective atoms of the electro-neutral residues from the Cornell et al. force
field6. The charges and atom types for both fragments are provided in the Supplementary
Table 7 and Supplementary Table 8.
Molecular dynamics simulation. The covalently bound alkyl-enzyme intermediates were
manually modeled from the enzyme-substrate complexes obtained from the docking
calculations. All enzyme-substrate complexes complying with the basic geometrical
prerequisites of SN2 reaction in the haloalkane dehalogenases10
were selected (Supplementary
Fig. 9-11), and the alkyl-enzyme intermediates were prepared by deleting the chloride atom and
placing the carbon atom 1 Å closer to the nucleophile oxygen atom of the aspartic acid residue
106 (DhaA and DhaA12) and 103 (DbjA). Water molecules from the crystal structure were
added to the systems to their original positions with the exception of water molecules
overlapping with the docked probes. The systems were neutralized by adding 16, 21 and 7 Na+
ions to DhaA, DhaA12 and DbjA complexes, respectively, using Tleap module of
AmberTools11 (University of California, 2010). Using the same module, an octahedron of
TIP3P water molecules11
was added to the distance of 10 Å from any solute atom in the
systems. Energy minimization and MD simulations were carried out in PMEMD module of
AMBER11 (University of California, 2010) using ff99SB force field12
. The investigated
systems were minimized by 500 steps of steepest descent followed by 500 steps of conjugate
gradient in five rounds of decreasing harmonic restraints. The restraints were applied as
follows: 500 kcal.mol-1
.Å-2
on all heavy atoms of protein, and then 500, 125, 25 and
0 kcal.mol-1
.Å-2
on backbone atoms only. The consequent MD simulations employed periodic
boundary conditions, the particle mesh Ewald method for treatment of the electrostatic
interactions13,14
, 10 Å cutoff for nonbonded interactions, and 2 ft time step with the SHAKE
algorithm to fix all bonds containing hydrogens15
. Equilibration simulations consisted of two
steps: (i) 20 ps of gradual heating from 0 to 300 K under constant volume, using a Langevin
Nature Chemical Biology: doi:10.1038/nchembio.1502
5
thermostat with collision frequency of 1.0 ps-1
, and with harmonic restraints of
5.0 kcal.mol-1
.Å-2
on the position of all protein atoms, and (ii) 2000 ps of unrestrained MD at
300 K using the Langevin thermostat, and constant pressure of 1.0 bar using pressure coupling
constant of 1.0 ps. Finally, production MD simulations were run for 50 ns with the same
settings as the second step of equilibration MD. Coordinates were saved in 5 ps interval, and
the trajectories were analyzed using Ptraj and Cpptraj modules of AmberTools11, and
visualized in Pymol 1.4.1 and VMD 1.8.916
. The average binding energies of each binding
mode were evaluated using Molecular-Mechanics/Generalized-Born Surface Area17,18
. Every
second frame from the equilibrated part of the molecular dynamics trajectories was selected for
the analysis. The following settings were used for the calculation: PBradii were set to mbondi2
and Generalized-Born model to 518
. The analysis was performed with MMPBSA.py python
script of AmberTools11. The binding modes with the lowest energies were then selected for the
detailed analysis of each system (Supplementary Table 9).
Nature Chemical Biology: doi:10.1038/nchembio.1502
6
Supplementary results
Supplementary Figure 1. Circular dichroism spectra of studied enzymes. All enzymes except
DhaA07 and DhaA09 exhibited circular dichroism spectra typical of folded /-proteins with
a predominately -helical content. Collected circular dichroism data are expressed in terms of
the mean residue ellipticity (ΘMRE).
Nature Chemical Biology: doi:10.1038/nchembio.1502
7
a b
Supplementary Figure 2. Comparison of the residues involved in the transplantation.
(a) Active site and tunnel lining residues variable between DbjA (red) and DhaA (magenta).
(b) Residues transplanted from DbjA (red) to DhaA12 (blue). EBR fragment is show as ribbon.
Nature Chemical Biology: doi:10.1038/nchembio.1502
8
Supplementary Figure 3. Superposition of the catalytic pentads. Pentads of DhaA12 (in blue)
and DbjA (in red) are represented by sticks. The 2Fo-Fc electron density map of the DhaA12
pentad contoured at 1σ represented by mesh.
Nature Chemical Biology: doi:10.1038/nchembio.1502
9
Supplementary Figure 4. MALDI-TOF MS spectra of DhaA12 and DhaA12-coumarin
complex. Black spectrum of the free enzyme (peak 1) is compared with red spectrum of the
enzyme-probe complex (peak 2). Seven consecutive mass spectra were recorded for each
sample and average mass values were calculated (RSD ≤ 0.1%). Difference of average masses
of enzymes and enzyme-probe complex corresponds to the mass of the coumarin probe. The
proteins contain the point mutation in the catalytic histidine His283Phe.
Nature Chemical Biology: doi:10.1038/nchembio.1502
10
Supplementary Figure 5. Acrylamide quenching of coumarin fluorescence for the haloalkane
dehalogenase variants. Stern-Volmer plots for DbjA (■), DhaA (Δ) and DhaA12 (○) provided
Stern-Volmer constants of 0.29 ± 0.04 M-1
, 0.11 ± 0.02 M-1
and 0.23 ± 0.06 M-1
respectively.
The linearity of those graphs indicates that those enzymes are specifically labeled. The
accessibility of the coumarin dye to the acrylamide molecules is substantially higher in the case
of the mutants DbjA and DhaA12 than in the case of DhaA. Stern-Volmer constant for DbjA
and DhaA12 are comparable, confirming the fact that the tunnel architecture of these enzymes
is similar. The emission spectra recorded before and after the quenching procedure did not
show any differences for all herein investigated mutants further indicating that the coumarin
dye possesses single location within the protein scaffold. Three experiments were performed
for DhaA and DbjA, and duplicate measurements were carried out in the case of DhaA12. Data
represent mean values ± s.d.
Nature Chemical Biology: doi:10.1038/nchembio.1502
11
a
b
Supplementary Figure 6. Structural formula of the fluorescent coumarin probe. Unbound
form (a) and its alkyl-intermediate complex with aspartic acid (b).
Nature Chemical Biology: doi:10.1038/nchembio.1502
12
Supplementary Figure 7. Two lowest-energy conformations of the coumarin probe. Probe is
capped with NME residue (atoms 27-32) employed for the fragment parameterization. The
atoms are numbered in concert with Supplementary Table 7.
Nature Chemical Biology: doi:10.1038/nchembio.1502
13
Supplementary Figure 8. Three lowest-energy conformations of the covalent linker. Linker is
capped with ACE residue (atoms 59-64) covalently bound to the aspartic acid residue capped
with NME (atoms numbered 47-52) and ACE (atoms 53-58) residues employed for the
fragment parameterization. The atoms are numbered in concert with Supplementary Table 8.
Nature Chemical Biology: doi:10.1038/nchembio.1502
14
Supplementary Figure 9. Five reactive binding modes of the probe bound in DbjA. Probe is
visualized with yellow sticks and enzyme as grey cartoon. The catalytic nucleophile and the
two halide-stabilizing residues are in cyan sticks.
Nature Chemical Biology: doi:10.1038/nchembio.1502
15
Supplementary Figure 10. Three reactive binding modes of the probe bound in DhaA. Probe
is visualized with yellow sticks and enzyme as grey cartoon. The catalytic nucleophile and the
two halide-stabilizing residues are in cyan sticks.
Nature Chemical Biology: doi:10.1038/nchembio.1502
16
Supplementary Figure 11. Five reactive binding modes of the probe bound in DhaA12. Probe
is visualized with yellow sticks and enzyme as grey cartoon. The catalytic nucleophile and the
two halide-stabilizing residues are in cyan sticks.
Nature Chemical Biology: doi:10.1038/nchembio.1502
17
Supplementary Table 1. Characteristics of enzymes used in this study.
Enzyme Mutations Engineered region Tm (C)a Reference
DhaA None None 50.5 0.3 This study
DhaA
+H272F H272F Catalytic base NT
b
19
DhaA07 ERB fragment + W141F + P153A ERB fragment unfolded This study
DhaA08 ERB fragment + W141F + P153A +
F155A ERB fragment 35.0 0.8 This study
DhaA09 ERB fragment + W141F + P153A +
F155A + V256A Main access tunnel unfolded This study
DhaA10 ERB fragment + W141F + P153A +
F155A + V256A + A183V Main access tunnel 40.5 0.6 This study
DhaA11 ERB fragment + W141F + P153A +
F155A + V256A + A183V + C187G Main access tunnel 42.5 0.8 This study
DhaA12
ERB fragment + W141F + P153A +
F155A + V256A + A183V + C187G
+ G182R + K186G
Main access tunnel 44.6 0.2 This study
DhaA12
+H283F
ERB fragment + W141F + P153A +
F155A + V256A + A183V + C187G
+ G182R + K186G + H283F
Catalytic base NTb This study
DbjA None None 53.6 0.6 20,21
DbjA
+H280F H280F Catalytic base NT
b
19
ERB fragment relates to the insertion of His-His-Thr-Glu-Val-Ala-Glu-Glu-Gln-Asp-His between the residues 141
and 142 of DhaA wild type. Cumulatively added mutations are in bold. The point mutation in the catalytic base
(His272Phe, His283Phe and His280Phe) enables covalent binding of fluorescence probe required for
time-dependent fluorescence experiments. aTm – melting temperature; data represent mean values ± s.d. calculated
from triplicate experiments bNT – not tested
Nature Chemical Biology: doi:10.1038/nchembio.1502
18
Supplementary Table 2. Enantioselectivity of wild type and variants with bromoalkanes
and bromoester.
Enzyme E-value
2-Bromopentane 2-Bromohexane 2-Bromoheptane Ethyl 2-bromopropionate
DhaA 8 4 2 72
DhaA08 5 4 2 163
DhaA10 9 11 2 > 200
DhaA11 8 12 2 > 200
DhaA12 7 11 10 > 200
DbjAa 145 68 28 > 200
Data was fitted by nonlinear regression based on the simulation with the standard error of the fit
between 5 and 25%. a data from
20.
Nature Chemical Biology: doi:10.1038/nchembio.1502
19
Supplementary Table 3. Data collection and refinement
statistics for the crystal structure of DhaA12.
DhaA12
Data collection
Space group P212121
Cell dimensions
a, b, c (Å) 51.58, 68.69, 84.37
() 90.00, 90.00, 90.00
Resolution (Å) 31.39–1.78 (1.88-1.78)a
Rmerge 8.4 (37)
I / I 14.4 (4.2)
Completeness (%) 100.0 (99.9)
Redundancy 4.1 (4.1)
Refinement
Resolution (Å) 1.78
No. reflections 34,658
Rwork / Rfree 0.176/0.207
No. atoms
Protein 2,537
Chloride ion 1
Water 270
Non-H atoms 2,808
B-factors (Å2)
Protein 16.7
Ligand/ion 12.7
Water 24.5
R.m.s. deviations
Bond lengths (Å) 0.007
Bond angles () 0.024
Parameters were obtained from a single crystal of DhaA12 protein.
a The highest-resolution shell is shown in parentheses.
Nature Chemical Biology: doi:10.1038/nchembio.1502
20
Supplementary Table 4. Parameters characterizing the overall time-dependent fluorescence
shift.
Enzyme 0
(cm-1
)
(cm
-1)
1 (ns)
(A1)
2 (ns)
(A2)
3 (ns)
(A3) r (ns) % obs.
DhaA 23300±60 950±80 0.2
(310)
1.3
(340)
12.6
(200) 4.1±0.3 90±5
DhaA12 23300±70 1000±100 0.2
(400) -
9.1
(400) 3.8±0.4 85±6
DbjA 23400±60 1300±140 < 0.1
(280)
1.4
(220)
11.0
(370) 2.8±0.3 62±9
0 corresponds to the emission maximum of “time zero spectrum” which is a spectrum emitted from the
completely non-relaxed Franck-Condon state and which was determined by the modified “time 0
estimation” as described previously19
. The rest of the given parameters was obtained by the analysis of the
time dependence of time resolved emission spectra (TRES): ) stands for the emission maximum of the
spectrum from fully relaxed state which was obtained by extrapolation of (t) to t . represents the
overall dynamic Stokes Shift derived as 0 - ultiexponential fitting of time evolution of the
TRES peak maximum (t), and analysis of the autocorrelation curve C(t): i are the fitted relaxation times,
Ai correspond to the amplitudes, r is the integral relaxation time, and % obs. corresponds to the percentage
of the solvent relaxation process captured with the given experimental time resolution of 30 ps. Four
experiments were performed for DhaA and DbjA, and duplicate measurements were carried out in the case
of DhaA12. Data represent mean values ± s.d. i and Ai parameters were gained by the non-linear least
square fitting with 3-exponential function. Fitting procedure yielded standard deviation lower than 10% for
all the parameters.
Nature Chemical Biology: doi:10.1038/nchembio.1502
21
Supplementary Table 5. Parameters describing the dynamics and hydration of wild types
and DhaA12 mutant.
a data from time-dependent fluorescence shift experiments; data represent mean values ± s.d. calculated from
four experiments for DhaA and DbjA, and two measurements for DhaA12. b data from molecular dynamics
simulations; data represent mean values ± s.d. calculated from 3,000 snapshots.
Enzyme
Dynamics Hydration
Integral relaxation
time [ns]a
Integral B-factor of the
tunnel mouth near dye [Å2]
b
Dynamic Stokes
shift [cm-1
]a
Average number of water
molecules around dyeb
DhaA 4.1± 0.3 132 950±80 6±2
DhaA12 3.8± 0.4 217 1000±100 6±2
DbjA 2.8± 0.3 315 1300±140 10±2
Nature Chemical Biology: doi:10.1038/nchembio.1502
22
Supplementary Table 6. Variable residues in the second and the third shell of
the active sites of DbjA and DhaA12.
Residues Location
DbjA DhaA12
Phe66 Met69 Second shell
Gln102 His105 Second shell
Thr106 Ser109 Second shell
Met132 Ile135 Second shell
Ile185 Val188 Second shell
Val210 Leu213 Second shell
Leu230 Val235 Second shell
Val255 Ile258 Second shell
Ile44 Leu47 Third shell
Ala101 Ile104 Third shell
Ala109 Gly112 Third shell
Phe125 Cys128 Third shell
Ala233 Tyr236 Third shell
His234 Met237 Third shell
Ser256 Pro259 Third shell
Leu275 Ile278 Third shell
Nature Chemical Biology: doi:10.1038/nchembio.1502
23
Supplementary Table 7. Atom types and charges for the fluorescent probe used in the
simulations.
Atom number Atom type Charge
1 N2 -0.8691
2 H 0.3788
3 H 0.3788
4 CA 0.3555
5 CA -0.2906
6 HA 0.1770
7 CA -0.1881
8 HA 0.1741
9 CA 0.0162
10 CA 0.1939
11 CA -0.2725
12 HA 0.1773
13 OA -0.3635
14 CA 0.7418
15 O -0.5652
16 CA 0.0595
17 CT -0.1794
18 HC 0.0799
19 HC 0.0799
20 HC 0.0799
21 CA -0.1896
22 CT -0.0634
23 HC 0.0502
24 HC 0.0502
25 C 0.5745
26 O -0.5863
Atom numbers are defined in Supplementary Fig. 7. Atom types accordingly to Cornell et al. force field6.
Nature Chemical Biology: doi:10.1038/nchembio.1502
24
Supplementary Table 8. Atom types and charges for the covalent linker bound to the aspartic
acid residue used in the simulations.
Atom number Atom type Charge
1 N -0.4520
2 H 0.2592
3 CT 0.0703
4 H1 0.0632
5 H1 0.0632
6 CT 0.1685
7 H1 0.0333
8 H1 0.0333
9 OS -0.4819
10 CT 0.1700
11 H1 0.0315
12 H1 0.0315
13 CT 0.1970
14 H1 0.0266
15 H1 0.0266
16 OS -0.4681
17 CT 0.1421
18 H1 0.0196
19 H1 0.0196
20 CT 0.0397
21 HC 0.0104
22 HC 0.0104
23 CT 0.0190
24 HC -0.0126
25 HC -0.0126
26 CT 0.0131
27 HC -0.0046
28 HC -0.0046
29 CT 0.0142
30 HC 0.0252
31 HC 0.0252
32 CT 0.0353
33 H1 0.0768
34 H1 0.0768
35 OS -0.3987
36 C 0.7481
37 O -0.5573
38 CT -0.2840
39 HC 0.0969
40 HC 0.0969
41 CT 0.1284
42 H1 0.0192
43 N -0.4156
44 H 0.2718
45 C 0.5972
46 O -0.5679
Atom numbers are defined in Supplementary Fig. 8. Atom types accordingly to Cornell et al. force field6.
Nature Chemical Biology: doi:10.1038/nchembio.1502
25
Supplementary Table 9. Binding energies of individual binding modes of the fluorescent
probe.
Enzyme Binding mode
Enzyme-probe binding energy [kcal/mol]
DbjA DbjA_18 DbjA_21 DbjA_28 DbjA_30 DbjA_38
-13±3 -11±2 -10±2 -19±2 -11±2
DhaA DhaA_4 DhaA_10 DhaA_33
- - -16±3 -21±2 -11±2
DhaA12 DhaA12_4 DhaA12_28 DhaA12_40 DhaA12_45 DhaA12_65
-23±3 -9±3 -9±3 -11±2 -8±1
Individual binding modes are depicted in Supplementary Fig. 9-11. Binding modes selected for further analysis are
in bold. Data represent mean values ± s.d. calculated from 3,000 snapshots.
Nature Chemical Biology: doi:10.1038/nchembio.1502
26
Supplementary references
1. Gordon, J. C. et al. Nucleic Acids Res. 33, W368–371 (2005).
2. Dewar, M. J. S., Zoebisch, E. G., Healy, E. F. & Stewart, J. J. P. J. Am. Chem. Soc. 107,
3902–3909 (1985).
3. Sanner, M. F. J. Mol. Graph. Model. 17, 57–61 (1999).
4. Morris, G. M. et al. J. Comput. Chem. 19, 1639–1662 (1998).
5. Solis, F. J. & Wets, R. J. B. Math. Oper. Res. 6, 19–30 (1981).
6. Cornell, W. D. et al. J. Am. Chem. Soc. 117, 5179–5197 (1995).
7. VanBeek, D. B., Zwier, M. C., Shorb, J. M. & Krueger, B. P. Biophys. J. 92, 4168–4178
(2007).
8. Dupradeau, F.-Y. et al. Phys. Chem. Chem. Phys. 12, 7821–7839 (2010).
9. Vanquelef, E. et al. Nucleic Acids Res. 39, W511–517 (2011).
10. Hur, S., Kahn, K. & Bruice, T. C. Proc. Natl. Acad. Sci. U. S. A. 100, 2215–2219 (2003).
11. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. J. Chem.
Phys. 79, 926–935 (1983).
12. Hornak, V. et al. Proteins 65, 712–25 (2006).
13. Darden, T., York, D. & Pedersen, L. J. Chem. Phys. 98, 10089–10092 (1993).
14. Essmann, U. et al. J. Chem. Phys. 103, 8577–8593 (1995).
15. Ryckaert, J.-P., Ciccotti, G. & Berendsen, H. J. C. J. Comput. Phys. 23, 327–341 (1977).
16. Humphrey, W., Dalke, A. & Schulten, K. J. Mol. Graph. 14, 33–8, 27–8 (1996).
17. Still, C., Tempczyk, A., Hawley, R. & Hendrickson, T. J. Am. Chem. Soc. 112, 6127–6129
(1990).
18. Onufriev, A., Bashford, D. & Case, D. A. Proteins 55, 383–394 (2004).
19. Jesenská, A. et al. J. Am. Chem. Soc. 131, 494–501 (2009).
20. Prokop, Z. et al. Angew. Chem. Int. Ed Engl. 49, 6111–6115 (2010).
21. Chaloupkova, R., Prokop, Z., Sato, Y., Nagata, Y. & Damborsky, J. FEBS J. 278, 2728–
2738 (2011).
Nature Chemical Biology: doi:10.1038/nchembio.1502